rubenada commented on issue #1684: [CALCITE-3627] Incorrect null semantic for ROW function URL: https://github.com/apache/calcite/pull/1684#issuecomment-570529739 @DonnyZone I think you are in the right track when you say that the nullability of `ROW` function is not the same as the nullability of a composite type; perhaps @zabetak could confirm, I think he knows the SQL standard better than myself :) Apart from that, I agree with you, `NullPolicy` for `ROW` function should not be ANY, so the current patch LGTM. If possible, I would add more tests (maybe in struct.iq?) to verify the different ROW nullability scenarios, e.g. (expected values taken from pg): ``` select row(null, null, null) is null -- expected true select row(null, null, null) is not null -- expected false select row(null, 1, null) is null -- expected false select row(null, 1, null) is not null -- expected false select not( row(null, 1, null) is null ) -- expected true ```
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
